[torquedev] CMake'ing Torque

Ken Nielson knielson at adaptivecomputing.com
Wed Apr 7 09:00:31 MDT 2010


Baer, Troy wrote:
> On Wed, 2010-04-07 at 15:14 +0200, Simon Toth wrote: 
>   
>> Dne 7.4.2010 04:25, Garrick Staples napsal(a):
>>     
>>> I still don't understand why anyone would want to convert an actively
>>> maintained autoconf setup that runs well on every platform.
>>>       
>> The reasons are many:
>>
>> * cmake is much easier to maintain/modify/use
>> * cmake does handle most of the logic internally (no need to write
>> custom scripts, and crazy make rules)
>> * cmake runs pretty much everywhere (including platforms not supported
>> by automake)
>> * cmake is a lot faster (this does matter, only when you always compile
>> from scratch)
>> * cmake is just a part of a larger ecosystem (ctest,cdash,cpack)
>> * cmake is used by many opensource projects and is very actively developed
>> * transformation from automake to cmake is relatively painless
>> (transformation scripts are avaiable)
>>     
>
> These are all good reasons why a new project might adopt cmake.
> However, TORQUE is *not* a new project, and it has an existing build
> system that works.  Moreover, I would guess that most of the active
> TORQUE developers are either not at all familiar with cmake or more
> familiar with autoconf than cmake.  You are in effect asking the TORQUE
> dev team to support an entirely new build system, with which many of
> them are not familiar, in addition to the existing one.  Unless you're
> volunteering to be the person who supports cmake in TORQUE, it is going
> to be difficult to convince Garrick, et al. to do this IMHO.
>
> 	--Troy
>   

I echo Troy and Garrick's comments. This is a legacy product and big 
changes to how things are done can cause a great deal of pain to those 
already using it.

But before we dismiss Dominique I would like to see the work he has done 
so everyone can see what it does.

Ken Nielson
Adaptive Computing


More information about the torquedev mailing list